Jump to content

Sign in to follow this  
Freemorpheme

Hard surface robot questions

Recommended Posts

13 hours ago, Freemorpheme said:

Can you think why the squares have been rounded off by the SDS even though I filled them with support loops?

 

Yes. Because there is nothing supporting the corners - you have only supported them in one direction. And there is not enough polys to work with to do borrowed edge corner controls. The only way you could fix that without more polys is by SDS edge weighting the corners to something like 80%.

 

This is what I did in my testings...

 

1. Flat Plane with helpful amount of segments (I went for 53; ((6 x 8 sections + the diagonals)), so that the top indented parts will be even and match at either end. 

2. Spline wrapped an instance of that (circle as source), and adjusted original's length to scale it to fit that spline.

3. Began my modellings in, thusly, working on the bottom plane, and viewing live results above in the wrapped version...

 

Robot Wrap 1.jpg

 

4. Then, put the instance inside an L1 OpenSubDiv SDS, and moved that into a group with the wrap, so it wrapped that instead with double the resolution.

5. Then took an editable copy of that to which I added thickness with extrude, then extruded in the cutout waist bits. 

 

If we don't apply further subdivision, then this might be OK remaining here as a hard surface like the one on the left below.

 

Robot Wrap 2.jpg

 

Or we can use SDS edge weighting and slide some loops about slightly to sharpen corners under 2 further levels of OpenSubDiv OR Catmull Clark SDS like the one on the right...

 

Either looks OK for me, but will depend on how close you're getting and how sharp you need those corners to be...

 

Hope that helps...

 

CBR

Share this post


Link to post
  • Topic Author
  • 10 hours ago, Cerbera said:

    Hope that helps...

     

    Thanks. You didnt happen to keep the C4D file, did you? I have remade all that and it is subdividing wildly differently from yours. I am doing something fundamentally wrong. You appear to have virtually no control loops in the two little boxes and they are dividing beautifully. I have a million loops in and they are wrong as balls.

    Share this post


    Link to post
    33 minutes ago, Freemorpheme said:

     

    I am doing something fundamentally wrong. You appear to have virtually no control loops in the two little boxes and they are dividing beautifully. I have a million loops in and they are wrong as balls.

     

    Have you done the corner weighting on these highlighted edges particularly ? And are you using L2 OpenSubDiv rather than regular Catmull Clark ? It subdivides subtly differently and is marginally better at reacting to edge weighting.

     

    image.png

     

    If that's not it, feel free to send me your file, and I'll see what's going wrong if I get some time later this afternoon...

     

    CBR

    Share this post


    Link to post
    12 hours ago, Cerbera said:

     

    Hope that helps...

     

    CBR

     

    Such a great idea modelling it flat then spline wrapping it into a cylinder shape afterwards. 

    Share this post


    Link to post
  • Topic Author
  • OK, after a mere 7 years of practice and 4 hours of work I have made.... this...

     

    Capture.thumb.JPG.3e9d145eae25e0d3e8ff1c05e76b970e.JPG

     

    I challenge anyone to come up with a more perfect piece of geometry...  god help me...

     

    Is this OK? I really tried hard!

    Share this post


    Link to post

    Maybe could be adjusted polygon flow into loops as posible, but if not some artefacts in rendered mesh, it could be OK.

    fast try...

    polyflow_1.thumb.png.93e19eac5e30cd9b5660b9dff9fd9511.png

    polyflow_2.thumb.png.922e81947485a55ee7a53bd55f419305.png

     

    Ofcourse, everything is dependent on whole mesh, where the loops go etc., if could be added some extra loop etc...

    ...there is many if... 🙂

    Share this post


    Link to post
  • Topic Author
  • 11 minutes ago, bezo said:

    fast try

    😞

     

    I know after many years it must be difficult to analyse how you do it, but can you give me an insight into how you think about making edge flow and quads? Do you just see it immediately or do you mess about with the mesh a bit? When you look at my mesh, do you instantly see what is wrong?

    Share this post


    Link to post
    On 12/7/2019 at 10:47 PM, Freemorpheme said:

    Do you just see it immediately or do you mess about with the mesh a bit?

     

    In my case, it's a little from column A, and a little from column B 🙂

     

    When you are very experienced, you tend to intrinsically know what will cause SDS distortion and what won't. Also the basic tenets and rules of modelling are always at the forefront of mind. If it helps, my thought process when first seeing your robot was something like what follows...

     

    • Oh, it's a cylinder, therefore a curved surface, therefore we need to be very careful to have enough loops that define that radial curvature, so that when we add details that curvature is not disturbed..
    • The details we need to put in place are very much smaller than the overall shape, so we need more segmentation (than just what is required to merely hold a cylindrical shape) vertically and radially because of that.
    • The shapes we need to cut in need to have very sharp corners, which instantly makes this much more challenging, because if we use control loops for that corner sharpening we risk a lot of variation in polygon density, which in turn risks SDS deformation when that is used with a smooth curved form. Mental note made to try SDS edge weighting instead of control loops, or to try and get the control another way (like via applied subdivision).
    • What sort of poly density do I need in my basic cylinder to be able to cut in those details ?  
    • Some of the details are at 45 (ish) degrees to the main edge flow (which goes radially around, and straight up and down). That is potentially a big problem because edges at 45 degrees to flow on a curved surface are highly likely to distort under SDS, so we must minimize those as much as possible, and surround them closely (but evenly) with straight form topology, so the effects of the angled details are constrained by the surroundings.
    • It is much easier to work on a flat mesh than it is a curved one (axis remains correct) , so can we start like that and add the curvature later ? Yes we could in this case.
    • Does that have any additional advantages ? Yes. It means that we can work on a fairly mid-poly form, and if that doesn't wrap nicely into a cylinder we will also have the option of putting the base model under SDS, which will subdivide it, and wrapping that instead, which will double or quadruple our resolution for free, and surprisingly often saves the day when cutting awkward details into curvy meshes.
    • Let's base our initial radial segment count on the rim details at the top. If we make the diagonal lines 1 poly wide, that tells us roughly how big each should be, and tells me what segmentation I should give the plane object I start this with.

    So those were my initial considerations, which gave me enough direction to decide on the spline wrap method, and to know what poly density I need to start.

     

    hs start.jpg

     

     

    This is my second iteration of the mesh, so is better than my first attempt. That's usually true in fact - the more times you build the same thing the better your results get. But, as you can hopefully see, in the lower plane (the unsubdivided base mesh) even this much poly density isn't enough to give sharp corners on those details and angled seams.

     

    But, by using edge weighting we can circumvent that and just avoiding having to add any corner control loops. I simply selected these edges, and set their edge weighting to 80%.

     

    image.png

     

    Now see the subdivided result above (it is this I will be wrapping), and notice that all the corners where I set that weighting to remain hard under SDS. Now we can avoid any contra-flow loops and variation in poly density that would be likely to cause problems when this is wrapped cylindrically. This is arguably better than yours or Bezo's approach because in my version there are almost no edges in contra-flow, which means that when curved this geo should provide a distortion free result, with or without additional SDS. And so it does, as we can see here...

     

    image.png

     

    Here is that start file so you can really look into it. Was made in R21, but hopefully should open in 17.

     

    Robot Wrap Start.c4d

     

    CBR

    Share this post


    Link to post
    18 hours ago, Freemorpheme said:

    😞

     

    I know after many years it must be difficult to analyse how you do it, but can you give me an insight into how you think about making edge flow and quads? Do you just see it immediately or do you mess about with the mesh a bit? When you look at my mesh, do you instantly see what is wrong?

     

    It's not that difficult once you understand a number of basic rules. I would recommend this book:

    http://pushingpoints.com/v2/the-pushing-points-topology-workbook/

     

    The second book dives into character head poly flow. But the first book should clear up any misunderstandings about poly flow.

     

    Share this post


    Link to post
  • Topic Author
  • On 12/8/2019 at 2:49 PM, Cerbera said:

    those were my initial considerations

    Thanks, as always for your detailed response. It's invaluable to imagine the thinking behind, as well as to see the mesh. For instance I see now you can split a quad into three quads, who knew? I will plod on and report back with my next roadblock...

    Share this post


    Link to post

    Join the conversation

    You can post now and register later. If you have an account, sign in now to post with your account.

    Guest
    Reply to this topic...

    ×   Pasted as rich text.   Paste as plain text instead

      Only 75 emoji are allowed.

    ×   Your link has been automatically embedded.   Display as a link instead

    ×   Your previous content has been restored.   Clear editor

    ×   You cannot paste images directly. Upload or insert images from URL.

    Sign in to follow this  

    • Recently Browsing   0 members

      No registered users viewing this page.

    Latest Topics

    Latest Comments

    ×
    ×
    • Create New...